Re: File locking anomaly under 2.0.30

Matthias Urlichs (smurf@work.smurf.noris.de)
25 Jul 1997 17:32:29 +0200


alan@lxorguk.ukuu.org.uk (Alan Cox) writes:
>
> I think its so obscure and changing core code that might burn others - we've
> had fascinating lock/unlock bugs in 2.0.14-18 for example that unless you
> can cite a POSIX.1/POSIX.4 reference on this it shouldnt be fixed in 2.0.x
>
You mean you want to leave the dangling fl_owner pointer in the lock
structure.

Sure, go ahead... :-/

> To start with there may be other people relying on the existing behaviour
> to wait for all threads of a task to exit...
>
I cannot believe that anybody would rely on that. My patch only frees the
dangling lock and allows the other threads to continue using the file.